The Future Of Human Intelligence And Artificial Intelligence

The field of artificial intelligence is very fascinating and fantastic. The development process is full of ups and downs. At each climax, celebrities predict that miracles will happen after a period of time, and then it is always disappointing but unable to give up. Such a story is the result of the development of big data, computing power and algorithms. Today, big data is no longer a problem, and computing power is also rapidly improving. A variety of quantum computing prototypes have been designed and can be put into commercial use in the near future.

Genius and algorithm, written by British mathematician Marcus Du sotoy, is a valuable popular science book which has a detailed understanding of the latest development of artificial intelligence algorithms. By analyzing the progress of algorithms in chess games, mathematical proof, painting, composition, content creation and translation, the author compares the similarities and differences between human thinking and AI mathematical thinking, so as to help readers objectively understand and evaluate the development path of AI technology and its impact on human beings.

One of the main controversies about whether AI can achieve human intelligence in the future is that the result of machine learning is a "black box", and human beings can not understand the principle on which intelligent machines correctly answer questions. Others insist that machines don't learn automatically, they can only learn and work under the arrangement of human beings, so AI can never catch up with human intelligence. These views have lagged behind the actual progress of AI. Taking alphago as an example, people originally thought that this kind of game with no fixed chess path was different from that of chess. AI was not competent. As a result, it defeated the nine section masters of go. The new version of alphago zero can become much more powerful than alphago in a few decades without learning any human combat data. This means that AI can start with nothing and defeat human experts by learning or training themselves. For many people, a few days' study can only reach the level of introduction to go.

It's not surprising that humans can't explain the behavior of AI based on machine learning. Think about it carefully. In fact, in many cases, we don't understand that we will think or do that. But humans are different in that they are good at "summarizing" and can make up a set of reasons to explain their thoughts or behaviors. And AI machines don't have the motivation to interact with humans. It's just that it doesn't speak, and people don't know what the "patterns" or the reasons are. In many cases, the reason or explanation of human identification is not correct, which is the development process of cognition. Another difference between human and computer is that human beings are always learning and recognizing patterns by using external input and their own perceptual information, which is the process of summarizing, reasoning and memorizing. People who are good at it seem to be knowledgeable or smarter than others. Machine learning actually adopts the same principle and process, but most AI learning is still in the passive stage. If machines, like humans, constantly accept external information, learn automatically, adjust feedback, and summarize, AI is obviously better than humans.

From the practical cases of AI in creativity provided in the book, it can be seen that one of the differences between intelligent machines and human thinking is that machines have no thinking blind areas that can not be touched. Because of the influence of millions of years of evolution, human brain has formed a thinking mode that must rely on emotion and motivation. This dependence has become a set of fixed patterns, which limits many areas that human can explore. AI algorithm out of the chess move, painting and music style, the formation of text content, will make people in surprise, recognize the possibility of being ignored by human beings, it can be said that human beings recognize their own limitations and shortcomings. This continuous breakthrough of the constraints of the original system and get infinite creativity beyond the human capacity.

Artificial intelligence algorithm has invaded the field that most depends on human intelligence. From the perspective of the author's mathematics major, there is no substantial obstacle to mathematical proof through algorithm. In recent years, the rapid development of artificial intelligence in the field of human expertise has begun to cause human deep concern. Artificial intelligence is not limited in composition, literature, painting and other aspects of creativity, can create a lot of surreal works. In the laboratory, researchers found that a new algorithm can make dynamic adjustments when interacting with data. After a period of time, even the programmers who originally developed it may not be able to understand the choice made by the algorithm. Intelligent machines can even form a mode of communication in which human beings have no legal solutions.

If machines and machines, or machines and humans can communicate, then there will be no distinction between humans and machines at the intellectual level, and machines will surpass humans in many ways. This is what Kurzweil calls the "singularity," and the miracle doesn't seem particularly far away. Human group behavior is restricted by ethics and law, while machines have neither emotion nor empathy, nor any objective environment to form norms. If human beings lose control of autonomous learning algorithms, can intelligent machines form behavior norms compatible with human society? Can human beings set higher-level restrictions on machine autonomous learning algorithm and draw a boundary of action that can not be overstepped by machines? With the rapid development of artificial intelligence algorithm, these problems become more and more important.

?